Nonreinforced substrates for use as embedded capacitors: a thin dielectric capacitive material shows promising electrical performance and is a drop-in substitute for FR-4. (Embedded Capacitance Substrates).

As CPUs increase in performance, the number of passive components on the surface of printed circuit boards is increasing dramatically. To reduce the number of components and improve electrical performance (e.g., reduce inductance), designers embed capacitive layers in the PCB. The majority of products use reinforced epoxy laminates. These are relatively easy to handle, but the thickness and Dk limit the effectiveness of the layer to perform as a capacitor. Other materials being developed are thinner (and thus increase capacitance), but either have problems with dielectric breakdown strength, handling or only marginal improvement over the existing material.

Here a novel nonreinforced substrate is reviewed, as are the material selection process, substrate processing and electrical performance. Standard capacitive material is typically 50 [micro]m (0.002") thick dielectric material, mostly for telecom and networking applications (1). For this high-end application, embedded capacitor technology has been used to distribute capacitance, enhance signal integrity, reduce impedance at high frequency and dampen noise.

A number of published papers discuss materials for embedded capacitors. From an electrical performance standpoint, demand is increasing for thinner capacitor material--as thin as 10 [micro]m--as the signal frequency increases (2).

In this article, considerations of copper foil and dielectric resin properties (which make up the capacitor material) are discussed. Also, fabrication of the thin film capacitor material and its subsequent electrical performance are described.

Embedded capacitance materials are constructed from two metal layers (usually copper foils) with a dielectric polymer film layer in between. In order to design the construction of the capacitance material, properties of the copper foil and dielectric layer are the keys to determining the performance of capacitance materials. We studied the influences of copper foil, dielectric type and dielectric thickness to design the construction of the thin capacitance material.

Five types of copper foils were prepared to investigate the affect of foil on capacitance material performance. Their properties are listed in TABLE 1. The copper foil is expected to have good performance in terms of peel strength, capacitance value and insulation resistance.

FIGURE 1 shows the results of peel strength tests on various copper foils with a standard epoxy resin system. It was found that the higher the profile of the copper foils, the higher the peel strength. The absolute value of the peel strength should be different depending upon resin properties (e.g., toughness) of the dielectric, but the trend should be the same. The appropriate profile of copper foil must be selected, since an excessively high profile will result in a drop in production yield, due to shorts (especially when dielectric thickness gets thinner).

[FIGURE 1 OMITTED]

Capacitance using various types of copper foils was measured. Measured capacitance specimens were prepared by controlling the distance of the peak-to-peak distance of two sheets of copper foil to 20 [micro]m. Capacitance measurement results are shown in FIGURE 2. Capacitance values are indicated in the index, defining the capacitance of foil "E" as 100. It was found that capacitance value increases with the increase in profile of copper foil. This can be explained by the increase in surface area with the copper foil with higher profile.

[FIGURE 2 OMITTED]

One of the challenges for thin capacitive material is the physical property of the dielectric. The thin dielectric must be tough and flexible in order to withstand processing. Another challenge for thin dielectric is the insulation reliability, such as hi-pot (high potential) test and electromigration.

Two types of dielectric were selected for the evaluation. One (A) is a conventional, industry standard epoxy resin system. Another (B) is a proprietary, modified resin system, designed for capacitive material. With these two types of dielectric systems, substrates were prepared to evaluate physical properties and insulation reliability.

Resin toughness is an important property in determining the material's processability. Because the resin is so thin, if it were too weak it would break apart during etching. As shown in TABLE 2, resin type B exhibited about 4.5 times higher tensile strength than A, which signifies the advantage of B in withstanding PCB manufacturing processes.

Thinner dielectric layers raise concerns over the insulation reliability between the layers. Electromigration tests were conducted using standard conditions of 85[degrees]C/85%/35V, for two different resin systems (TABLE 3). The dielectric thickness was prepared at 10 [micro]m. Resin type A lasted 682 hours before electrically shorting. Resin type B lasted over 1,000 hrs. The differences in ionic contamination, moisture absorption and polymer structure seem to influence electromigration endurance.

Material Tests

Based on these investigations, a suitable copper foil and resin type was selected to construct and manufacture capacitive material of 10 [micro]m and 25 [micro]m dielectric thickness. Characteristics of the developed capacitive material are listed in TABLE 4.

Fabrication testing was conducted on thin film (10 and 25 [micro]m dielectric thickness) capacitance materials to validate the ability to process the material using conventional PCB manufacturing equipment. FIGURE 3 shows the patterned capacitance material. The material was processed at two PCB shops using the following processes.

1. Chemical etching.

2. Dry-film lamination.

3. Image transfer.

4. Pattern etching (both sides).

5. Black oxidize.

[FIGURE 3 OMITTED]

Due to the tough and flexible nature of the resin, the material was processed with no damage. Although processing thin capacitance material can be challenging and may require modifications to existing processes, the material was processable using conventional manufacturing steps. The patterned laminate was subjected to hi-pot testing. Every panel, including the 10 [micro]m-thick material, passed the 500 V test.

Scaling is a very important parameter for multilayer lamination. FIGURE 4 shows the measured result of the movement between the holes during processing. The novel material's movement was equal to that of 50 [micro]m core material. Hence, the scaling factor for the novel material can be the same as that of the 50 [micro]m core material.

[FIGURE 4 OMITTED]

Figures 5 to 7 show cross-sections of boards using the novel material. FIGURE 5 is a cross-section using 50 [micro]m dielectric thickness capacitive material as a reference, while FIGURE 6 uses 25 [micro]m and FIGURE 7 uses 10 [micro]m. Good bonding of the plated copper to the thin dielectric layers was observed. Compatibility with FR-4 laminates was good. Many electrical performance advantages, such as power distribution and electromagnetic interference (EMI), can be expected when thin capacitive material is used.

[FIGURES 5-7 OMITTED]

Conducted emissions on the supply line for an MPU operating at 40 MHz were measured using the Verfahren Gaseous Emission (VDE) method. No discrete decoupling capacitors were mounted on the four-layer board (FIGURE 8). Comparisons of conducted emissions between the standard 400 [micro]m laminate core and 10 [micro]m thin film capacitive core are shown in FIGURE 9. Significant reduction of emissions in the frequency range of 150 to 550 MHz, known as a difficult range to reduce when using discrete capacitors, was observed by the thin film capacitive core. Lower effective inductance and larger capacitance of the thin film core structure improves the performance of supply decoupling for MCUs.

[FIGURES 8-9 OMITTED]

The change in capacitance (due to change in dielectric constant) with change in frequency was also measured. FIGURE 10 shows that the material is relatively stable at these frequencies.

[FIGURE 10 OMITTED]

The novel material demonstrated improved electrical performance and capacitance values in conventional FR-4, and the ability to withstand 500 V hi-pot testing (even at 10 [micro]m) has not been demonstrated by any other thin polymer capacitive substrate. Further testing of capacitance vs. frequency and temperature changes is planned.

TABLE 1. Copper Foil Profile Properties

PROPERTIES                               FOIL TYPE

                      A * (1)   B * (2)   C * (3)   D * (4)   E * (5)

Copper weight         1oz       1oz       1oz       1oz       1oz
Profile Rz [micro]m   6.0       3.5       2.5       2.0       1.5

(1.) Standard HTE electrodeposited (ED) copper foil.

(2.) RTF (reverse-treated foil).

(3.) Super low profile ED foil.

(4.) Ultra low profile ED foil,

(5.) Wrought foil.

TABLE 2. Capacitance Measurement Results

RESIN TYPE                          A     B

Tensile Strength * (N/[mm.sup.2])   99   448

* Measured on 10 [micro]m thick resin

TABLE 3. Electromigration Test Results

RESIN TYPE       LENGTH (HRS.)

              N=1     N=2     Avg.

A            658     706     682

B            >1000   >1000   >1000

TABLE 4. Dielectric Material Characteristics

PROPERTIES                        25 [micro]m MATERIAL

Copper foil type                            B
Dielectric type                             B
Copper weight, oz.                          1
Peel strength, kN/m                        1.0
Dielectric thickness, [micro]m               25
Capacitance at 1 MHz                       0.16
Dk at 1 MHz                                4.0
Df at 1 MHz                                0.02
Dielectric breakdown, V                    >500
Tg (DMA), [degrees]C                       >220
Electrical migration (1), hrs.            >1000
Solder float (5X@288[degrees]C)          Passed

PROPERTIES                        10 [micro]m MATERIAL

Copper foil type                            C
Dielectric type                             B
Copper weight, oz.                          1
Peel strength, kN/m                        1.0
Dielectric thickness, [micro]m              10
Capacitance at 1 MHz                       0.35
Dk at 1 MHz                                4.0
Df at 1 MHz                                0.02
Dielectric breakdown, V                    >500
Tg (DMA), [degrees]C                       >220
Electrical migration (1), hrs.            >1000
Solder float (5X@288[degrees]C)          Passed

(1) RH85%/85[degrees]C/35V

ACKNOWLEDGMENTS

The authors would like to acknowledge our colleagues at Mitsui Mining & Smelting Co. Ltd. and Oak-Mitsui for their work for preparing and evaluating the material. We appreciate Atsushi Nakamura and his team at Hitachi Ltd. for providing data and suggestions. We also appreciate the work by Dr. J. Dougherty and his team at Pennsylvania State University for conducting frequency response testing.

This article was first published in the IPC Annual Meeting Proceedings, October 2002, and is used with permission of the authors.

REFERENCES

(1.) Jeffrey Gotro et al, "An Improved Laminate for Embedded Capacitance Applications," IPC Printed Circuit Expo Proceedings, March 1999.

(2.) Istvan Novak, "Embedded Passive Electrical Characterization Results," IPC Printed Circuit Expo Proceedings, April 2000.

(3.) Courtesy Hitachi Ltd.

(4.) Courtesy Hitachi Ltd.

(5.) T. Yamamoto, K. Yamazaki, E Kuwako, "Next Generation Embedded Capacitance Material" Electronic Circuit World Convention Proceedings, October 2002.

FUJIO KUWAKO is president of Oak-MitsuiTechnologies and TAKUYA YAMAMOTO is marketing manager. KAZUHIRO YAMAZAKI is senior researcher at Mitsui Mining and Smelting. JOHN ANDRESAKIS is vice president of strategic technology at Oak-MitsuiTechnologies. He can be reached at [e-mail].